Airbnb - relation between houses’ rent & Airbnb revenue on SA2 level

Models - nlme

1 Selection

Minimum 25 rental properties. Minimum 10 Airbnbs. Per month.

summary(APM_sa2_cor$mean_airbnb)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   11.19   24.70   42.44   98.78   94.26 3050.81

summary(APM_sa2_cor$mean_apm)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   29.82  125.44  240.56  358.67  443.70 4483.07

summary(APM_sa2_cor$min_airbnb)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   10.00   15.00   25.00   66.14   62.00 1782.00

summary(APM_sa2_cor$min_apm)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    25.0   108.0   213.0   321.3   391.0  4235.0

2 Data

Houses only

2.1 Number of areas

length(unique(data$SA2_MAIN16))

## [1] 654

data %>% 
  select(SA2_MAIN16, STE_NAME16) %>% 
  distinct() %>% 
  group_by(STE_NAME16) %>% 
  summarize(areas = n())

## # A tibble: 7 x 2
##   STE_NAME16         areas
##   <chr>              <int>
## 1 New South Wales      216
## 2 Northern Territory     1
## 3 Queensland           130
## 4 South Australia       45
## 5 Tasmania              35
## 6 Victoria             142
## 7 Western Australia     85

2.2 Temporal trends

ggplot(data, aes(x = date, y = rent_50p_int)) +
  geom_line(aes(group = SA2_MAIN16), alpha = .5) +
  # geom_smooth(method = "lm", se = FALSE) + 
  geom_smooth(aes(weight = record_count)) +
  theme_minimal() + xlab("") + ylab("Rent") +
  facet_wrap(~STE_NAME16)

ggplot(data, aes(x = date, y = adr_50p_int)) +
  geom_line(aes(group = SA2_MAIN16), alpha = .5) +
  # geom_smooth(method = "lm", se = FALSE) + 
  geom_smooth(aes(weight = n_listings_int)) +
  theme_minimal() + xlab("") + ylab("AWR") +
  facet_wrap(~STE_NAME16)

2.3 Rent vs AWR

2.3.1 All 27 months

data %>% 
  filter(STE_NAME16 != "Northern Territory") %>%
  ggplot(aes(x = rent_50p_int, y = adr_50p_int)) +
  geom_point(aes(size = n_listings_int)) +
  # geom_smooth(aes(weight = n_listings_int), method = "lm", se = FALSE)
  geom_smooth(aes(weight = n_listings_int)) +
  theme_minimal() + xlab("Rent") + ylab("AWR") +
  facet_wrap(~STE_NAME16)

2.3.2 All areas by month

data %>% 
  filter(STE_NAME16 != "Northern Territory") %>%
  filter(STE_NAME16 != "Australian Capital Territory") %>%
  ggplot(aes(x = rent_50p_int, y = adr_50p_int)) +
  # geom_point(aes(size = n_listings_int, col = STE_NAME16)) +
  # geom_smooth(aes(weight = n_listings_int), method = "lm", se = FALSE) +
  geom_smooth(aes(weight = n_listings_int, col = STE_NAME16)) +
  facet_wrap(~time) +
  theme_minimal() + xlab("Rent") + ylab("AWR") +
  theme(strip.text.x = element_blank()) 

2.3.3 Last month

data %>% 
  filter(STE_NAME16 != "Northern Territory") %>%
  filter(time == 27) %>%
  ggplot(aes(x = rent_50p_int, y = adr_50p_int)) +
  geom_point(aes(size = n_listings_int)) +
  # geom_smooth(aes(weight = n_listings_int), method = "lm", se = FALSE)
  geom_smooth(aes(weight = n_listings_int)) +
  theme_minimal() + xlab("Rent") + ylab("AWR") +
  facet_wrap(~STE_NAME16)

2.4 Rent over time

# rent linear
data %>% 
  # filter(SA2_NAME16 %in% c("Sydney - Haymarket - The Rocks", "Margaret River")) %>%
  ggplot(aes(x = time, y = rent_50p_int/1000)) +
  # annotate("text", x = 0, y = 200, label = unique(data$SA2_NAME16)) +
  # geom_point(aes(size = record_count), alpha = 0.3) +
  geom_smooth(aes(group = interaction(SA2_NAME16, STE_NAME16),
                  weight = record_count_int,
                  col = STE_NAME16), 
              method = "lm", se = FALSE) +
  facet_wrap(STE_NAME16~SA2_NAME16) +
  theme_minimal() + xlab("") + ylab(("")) +
  theme(strip.text.x = element_blank()) 

# rent smooth
data %>% 
  # filter(SA2_NAME16 %in% c("Sydney - Haymarket - The Rocks", "Margaret River")) %>%
  ggplot(aes(x = time, y = rent_50p_int/1000)) +
  # annotate("text", x = 0, y = 200, label = unique(data$SA2_NAME16)) +
  # geom_point(aes(size = record_count), alpha = 0.3) +
  geom_smooth(aes(group = interaction(SA2_NAME16, STE_NAME16),
                  weight = n_listings_int,
                  col = STE_NAME16), se = FALSE) +
  facet_wrap(STE_NAME16~SA2_NAME16) +
  theme_void() +
  theme(strip.text.x = element_blank(),
        axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank(), 
        axis.title.y=element_blank(),
        axis.text.y=element_blank(),
        axis.ticks.y=element_blank())

2.5 AWR over time

## adr, linear
ggplot(data, aes(x = time, y = adr_50p_int/1000, group = SA2_NAME16)) +
  # annotate("text", x = 0, y = 200, label = unique(data$SA2_NAME16)) +
  # geom_point(aes(size = n_listings_int), alpha = 0.3) +
  geom_smooth(aes(group = interaction(SA2_NAME16, STE_NAME16),
                  col = STE_NAME16,
                  weight = n_listings_int),
              method = "lm", se = FALSE) +
  facet_wrap(STE_NAME16~SA2_NAME16) +
  theme_minimal() +
  theme(strip.text.x = element_blank()) 

# adr, smooth
ggplot(data, aes(x = time, y = adr_50p_int/1000, group = SA2_NAME16)) +
  # annotate("text", x = 0, y = 200, label = unique(data$SA2_NAME16)) +
  # geom_point(aes(size = n_listings_int), alpha = 0.3) +
  geom_smooth(aes(group = interaction(SA2_NAME16, STE_NAME16),
                  weight = n_listings_int, 
                  col = STE_NAME16), se = FALSE) +
  facet_wrap(STE_NAME16~SA2_NAME16) +
  theme_minimal() +
  theme(strip.text.x = element_blank(),
        axis.title.x=element_blank(),
        axis.text.x=element_blank(),
        axis.ticks.x=element_blank(), 
        axis.title.y=element_blank(),
        axis.text.y=element_blank(),
        axis.ticks.y=element_blank())

3 Models

3.1 random region, random state, no adr

m1 <- lme(rent_50p_int ~ time, 
          random = ~ 1 | STE_NAME16/SA2_MAIN16, 
          data = data, method = "ML")

# summary(m1)
# 
# data$m1_fit <- fitted(m1)
# 
# ggplot(data, aes(x = rent_50p_int)) +
#   geom_point(aes(y = m1_fit, size = n_listings_int), 
#              alpha = 0.3) +
#   geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
#   facet_wrap(~STE_NAME16) +
#   theme_minimal() + xlab("Rent") + ylab("Predicted") 

3.2 random region, random state, with adr

m2 <- lme(rent_50p_int ~ adr_50p_int + time, 
          random = ~ 1 | STE_NAME16/SA2_MAIN16, 
          data = data, method = "ML")

# summary(m2)
# 
# # anova(m1, m2)
# 
# data$m2_fit <- fitted(m2)
# 
# ggplot(data, aes(x = rent_50p_int)) +
#   geom_point(aes(y = m2_fit, size = n_listings_int), 
#              alpha = 0.3) +
#   geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
#   facet_wrap(~STE_NAME16) +
#   theme_minimal() + xlab("Rent") + ylab("Predicted") 
# 
# plot(ggpredict(m2, "time"))
# 
# plot(ggpredict(m2, terms = "adr_50p_int"))

3.3 random time region

m3 <- lme(rent_50p_int ~ adr_50p_int + time, 
          random = ~ time | STE_NAME16/SA2_MAIN16, 
          data = data, method = "ML")

# summary(m3)
# 
# # anova(m2, m3) 
# 
# data$m3_fit <- fitted(m3)
# 
# ggplot(data, aes(x = rent_50p_int)) +
#   geom_point(aes(y = m3_fit, size = n_listings_int), 
#              alpha = 0.3) +
#   geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
#   facet_wrap(~STE_NAME16) +
#   theme_minimal() + xlab("Rent") + ylab("Predicted") 
# 
# plot(ggpredict(m3, "time"))
# # plot(ggpredict(m3, terms = c("time", "STE_NAME16")))
# 
# plot(ggpredict(m3, terms = "adr_50p_int"))
# # plot(ggpredict(m3, terms = c("adr_50p_int", "STE_NAME16")))

3.4 random region AR1

m4 <- lme(rent_50p_int ~ adr_50p_int + time, 
          random = ~ 1 | STE_NAME16/SA2_MAIN16, 
          correlation = corAR1(), 
          data = data, method = "ML")

# summary(m4)
# 
# # anova(m3, m4) 
# 
# data$m4_fit <- fitted(m4)
# 
# ggplot(data, aes(x = rent_50p_int)) +
#   geom_point(aes(y = m1_fit, size = n_listings_int), 
#              alpha = 0.3) +
#   geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
#   facet_wrap(~STE_NAME16) +
#   theme_minimal() + xlab("Rent") + ylab("Predicted") 
# 
# plot(ggpredict(m4, "time"))
# 
# plot(ggpredict(m4, terms = "adr_50p_int"))

# fixef(m4)
# ranef(m4)
# plot(ACF(m4), alpha = 0.05)

3.5 random time region AR1

m5 <- lme(rent_50p_int ~ adr_50p_int + time, 
          random = ~ time | STE_NAME16/SA2_MAIN16, 
          correlation = corAR1(form = ~ time | STE_NAME16/SA2_MAIN16), 
          data = data, method = "ML")

# summary(m5)
# 
# # anova(m4, m5)
# # anova(m3, m5)
# # anova(m2, m5)
# 
# data$m5_fit <- fitted(m5)
# 
# ggplot(data, aes(x = rent_50p_int)) +
#   geom_point(aes(y = m1_fit, size = n_listings_int),
#              alpha = 0.3) +
#   geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
#   facet_wrap(~STE_NAME16) +
#   theme_minimal() + xlab("Rent") + ylab("Predicted")
# 
# plot(ggpredict(m5, "time"))
# 
# plot(ggpredict(m5, terms = "adr_50p_int"))

# fixef(m5)
# ranef(m5)
# plot(ACF(m5), alpha = 0.05)

3.6 Comparison

compare_performance(m1, m2, m3, m4, m5)

## Random effect variances not available. Returned R2 does not account for random effects.
## Random effect variances not available. Returned R2 does not account for random effects.
## Random effect variances not available. Returned R2 does not account for random effects.

##   Model Type      AIC      BIC     RMSE R2_conditional R2_marginal       ICC
## 1    m1  lme 157384.1 157423.0 17.55667             NA  0.15500976        NA
## 2    m2  lme 157361.4 157408.1 17.54794             NA  0.17508765        NA
## 3    m3  lme 148107.7 148185.5 12.37598      0.8510986  0.03601346 0.8455358
## 4    m4  lme 134155.2 134209.7 19.78898             NA  0.05358051        NA
## 5    m5  lme 133989.4 134074.9 17.04686             NA          NA        NA
##         BF
## 1       NA
## 2 1728.376
## 3      Inf
## 4      Inf
## 5      Inf

as_tibble(list(
  Model = as.character(c(1:5)),
  AIC = c(AIC(m1), AIC(m2), AIC(m3), AIC(m4), AIC(m5)))) %>% 
  ggplot() +
  geom_col(aes(y= AIC, x = Model)) +
  coord_flip() +
  theme_minimal()

3.7 Best model (AIC/BIC)

summary(m5)

## Linear mixed-effects model fit by maximum likelihood
##  Data: data 
##        AIC      BIC    logLik
##   133989.4 134074.9 -66983.68
## 
## Random effects:
##  Formula: ~time | STE_NAME16
##  Structure: General positive-definite, Log-Cholesky parametrization
##             StdDev     Corr  
## (Intercept) 89.8115025 (Intr)
## time         0.9317025 -0.149
## 
##  Formula: ~time | SA2_MAIN16 %in% STE_NAME16
##  Structure: General positive-definite, Log-Cholesky parametrization
##             StdDev      Corr  
## (Intercept) 205.4527220 (Intr)
## time          0.4813875 0.407 
## Residual     23.8781596       
## 
## Correlation Structure: AR(1)
##  Formula: ~time | STE_NAME16/SA2_MAIN16 
##  Parameter estimate(s):
##       Phi 
## 0.9189422 
## Fixed effects: rent_50p_int ~ adr_50p_int + time 
##                Value Std.Error    DF   t-value p-value
## (Intercept) 435.4568  37.44660 17002 11.628739  0.0000
## adr_50p_int   0.0024   0.00142 17002  1.711809  0.0870
## time          0.7566   0.37467 17002  2.019344  0.0435
##  Correlation: 
##             (Intr) ad_50_
## adr_50p_int -0.056       
## time        -0.132 -0.003
## 
## Standardized Within-Group Residuals:
##          Min           Q1          Med           Q3          Max 
## -10.33405353  -0.27431127  -0.02659564   0.24353926  10.05590881 
## 
## Number of Observations: 17658
## Number of Groups: 
##                 STE_NAME16 SA2_MAIN16 %in% STE_NAME16 
##                          7                        654

# anova(m4, m5)
# anova(m3, m5)
# anova(m2, m5)

# intervals(m5)

# plot_model(m5, type = "est")
plot_model(m5, type = "slope")

plot_model(m5, type = "resid")

plot_model(m5, type = "diag")

## [[1]]

## 
## [[2]]

## 
## [[3]]

data$m5_fit <- fitted(m5)

ggplot(data, aes(x = rent_50p_int)) +
  geom_point(aes(y = m5_fit, 
                 size = record_count),
                 # size = n_listings_int),
             alpha = 0.3) +
  geom_line(aes(y = rent_50p_int), col = "darkorchid4") +
  facet_wrap(~STE_NAME16) +
  theme_minimal() + xlab("Rent") + ylab("Predicted")

# plot(ggpredict(m5, "time"))
# 
# plot(ggpredict(m5, terms = "adr_50p_int"))

p1 <- plot(ggpredict(m5, "time"))

p2 <- plot(ggpredict(m5, terms = "adr_50p_int"))

plot_grid(p1, p2)

3.8 Prediction

3.8.1 WA

data %>% 
  filter(SA2_NAME16 %in% 
           c("Busselton", 
             "Margaret River", 
             "City Beach")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("Rent") + ylab("Fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

3.8.2 NSW

data %>% 
  filter(SA2_NAME16 %in% 
           c("Byron Bay", 
             "Avalon - Palm Beach")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

3.8.3 Qld

data %>% 
  filter(SA2_NAME16 %in% 
           c("Broadbeach Waters", 
             "Stanthorpe Region")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

3.9 ST outliers

data$diff <- (data$rent_50p_int - data$m5_fit) / data$rent_50p_int

3.9.1 Means from the whole time period

data %>% 
  group_by(SA2_NAME16) %>% 
  summarise(mean_diff = mean(diff),
            STE_NAME16 = first(STE_NAME16),
            n_listings_int = min(n_listings_int)) %>% 
  arrange(mean_diff) %>% 
  print(n = 50)

## # A tibble: 654 x 4
##    SA2_NAME16                           mean_diff STE_NAME16      n_listings_int
##    <chr>                                    <dbl> <chr>                    <int>
##  1 Flinders                               -0.0450 Victoria                   135
##  2 East Melbourne                         -0.0403 Victoria                    15
##  3 Bathurst Region                        -0.0349 New South Wales             19
##  4 Trigg - North Beach - Watermans Bay    -0.0326 Western Austra~             23
##  5 Colac Region                           -0.0275 Victoria                    19
##  6 Lenah Valley - Mount Stuart            -0.0246 Tasmania                    18
##  7 The Coorong                            -0.0242 South Australia             22
##  8 Floreat                                -0.0235 Western Austra~             14
##  9 Hamilton (Qld)                         -0.0200 Queensland                  10
## 10 Belmont - Ascot - Redcliffe            -0.0197 Western Austra~             34
## 11 Mandurah                               -0.0196 Western Austra~             45
## 12 Parkwood - Ferndale - Lynwood          -0.0193 Western Austra~             12
## 13 Bentley - Wilson - St James            -0.0193 Western Austra~             18
## 14 Hillarys                               -0.0190 Western Austra~             15
## 15 Samford Valley                         -0.0187 Queensland                  17
## 16 Orange Region                          -0.0187 New South Wales             11
## 17 Beckenham - Kenwick - Langford         -0.0183 Western Austra~             15
## 18 The Vines                              -0.0182 Western Austra~             10
## 19 Heathcote                              -0.0182 Victoria                    11
## 20 Falcon - Wannanup                      -0.0181 Western Austra~             43
## 21 Nambucca Heads Region                  -0.0180 New South Wales             27
## 22 Benalla                                -0.0180 Victoria                    17
## 23 Geraldton                              -0.0177 Western Austra~             15
## 24 Mandurah - South                       -0.0174 Western Austra~             10
## 25 Yorke Peninsula - South                -0.0171 South Australia             28
## 26 George Town                            -0.0169 Tasmania                    14
## 27 Victoria Park - Lathlain - Burswood    -0.0169 Western Austra~             52
## 28 Forster-Tuncurry Region                -0.0166 New South Wales            292
## 29 Wembley - West Leederville - Glenda~   -0.0163 Western Austra~             23
## 30 City Beach                             -0.0161 Western Austra~             13
## 31 Mandurah - East                        -0.0157 Western Austra~             29
## 32 Rockingham                             -0.0157 Western Austra~             23
## 33 Ellenbrook                             -0.0155 Western Austra~             13
## 34 Bayswater - Embleton - Bedford         -0.0151 Western Austra~             19
## 35 Coolangatta                            -0.0149 Queensland                  15
## 36 Safety Bay - Shoalwater                -0.0145 Western Austra~             25
## 37 Irwin                                  -0.0144 Western Austra~             13
## 38 Mount Nelson - Dynnyrne                -0.0141 Tasmania                    21
## 39 Freshwater - Brookvale                 -0.0141 New South Wales            171
## 40 Lakes Entrance                         -0.0140 Victoria                   104
## 41 Butler - Merriwa - Ridgewood           -0.0139 Western Austra~             11
## 42 Ballarat - South                       -0.0138 Victoria                    23
## 43 Tuart Hill - Joondanna                 -0.0138 Western Austra~             12
## 44 Queanbeyan Region                      -0.0137 New South Wales             11
## 45 Booragoon                              -0.0136 Western Austra~             30
## 46 East Victoria Park - Carlisle          -0.0136 Western Austra~             37
## 47 Mount Hawthorn - Leederville           -0.0135 Western Austra~             20
## 48 Maylands                               -0.0134 Western Austra~             17
## 49 Beechworth                             -0.0134 Victoria                    41
## 50 East Brisbane                          -0.0134 Queensland                  12
## # ... with 604 more rows

data %>% 
  filter(SA2_NAME16 %in% 
           c("Trigg - North Beach - Watermans Bay", 
             "Flinders")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

data %>% 
  filter(SA2_NAME16 %in% 
           c("Coolangatta", 
             "Hamilton (Qld)")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

data %>% 
  group_by(SA2_NAME16) %>% 
  summarise(mean_diff = mean(diff),
            STE_NAME16 = first(STE_NAME16),
            n_listings_int = min(n_listings_int)) %>% 
  arrange(desc(mean_diff)) %>% 
  print(n = 50)

## # A tibble: 654 x 4
##    SA2_NAME16                          mean_diff STE_NAME16       n_listings_int
##    <chr>                                   <dbl> <chr>                     <int>
##  1 Sydney - Haymarket - The Rocks         0.0431 New South Wales              32
##  2 Dover Heights                          0.0363 New South Wales              88
##  3 Toorak                                 0.0259 Victoria                     22
##  4 St Ives                                0.0225 New South Wales              12
##  5 Jindabyne - Berridale                  0.0219 New South Wales             124
##  6 Hobart                                 0.0215 Tasmania                    103
##  7 Randwick - North                       0.0213 New South Wales              61
##  8 Kuranda                                0.0201 Queensland                   19
##  9 Cronulla - Kurnell - Bundeena          0.0195 New South Wales              45
## 10 Broulee - Tomakin                      0.0192 New South Wales              42
## 11 St Leonards - Naremburn                0.0190 New South Wales              17
## 12 Paradise Point - Hollywell             0.0183 Queensland                   18
## 13 North Sydney - Lavender Bay            0.0174 New South Wales              20
## 14 Mascot - Eastlakes                     0.0171 New South Wales              13
## 15 Bondi Beach - North Bondi              0.0170 New South Wales             337
## 16 Bondi - Tamarama - Bronte              0.0161 New South Wales             280
## 17 Peregian Beach - Marcus Beach          0.0161 Queensland                   87
## 18 Kensington (NSW)                       0.0161 New South Wales              22
## 19 Kingsford                              0.0158 New South Wales              32
## 20 Lane Cove - Greenwich                  0.0158 New South Wales              57
## 21 Dulwich Hill - Lewisham                0.0157 New South Wales              28
## 22 Cremorne - Cammeray                    0.0156 New South Wales              54
## 23 Turramurra                             0.0153 New South Wales              13
## 24 Robertson - Fitzroy Falls              0.0152 New South Wales              38
## 25 Scottsdale - Bridport                  0.0149 Tasmania                     42
## 26 Prahran - Windsor                      0.0144 Victoria                     91
## 27 Bondi Junction - Waverly               0.0143 New South Wales             160
## 28 Paddington - Moore Park                0.0138 New South Wales             259
## 29 Southern Highlands                     0.0137 New South Wales              63
## 30 Frenchs Forest - Belrose               0.0136 New South Wales              33
## 31 Corrimal - Tarrawanna - Bellambi       0.0135 New South Wales              14
## 32 Haberfield - Summer Hill               0.0133 New South Wales              19
## 33 Chatswood (East) - Artarmon            0.0131 New South Wales              25
## 34 Mermaid Beach - Broadbeach             0.0131 Queensland                   54
## 35 Augusta                                0.0130 Western Austral~             89
## 36 Willoughby - Castle Cove - Northbr~    0.0129 New South Wales              61
## 37 Mosman Park - Peppermint Grove         0.0128 Western Austral~             31
## 38 Galston - Laughtondale                 0.0128 New South Wales              15
## 39 Brighton (Vic.)                        0.0125 Victoria                     60
## 40 New Town                               0.0122 Tasmania                     16
## 41 Box Hill North                         0.0119 Victoria                     14
## 42 Chatswood (West) - Lane Cove North     0.0118 New South Wales              19
## 43 Neutral Bay - Kirribilli               0.0117 New South Wales              41
## 44 Sussex Inlet - Berrara                 0.0117 New South Wales              38
## 45 Petersham - Stanmore                   0.0117 New South Wales              81
## 46 Marrickville                           0.0115 New South Wales              57
## 47 Yarra - North                          0.0115 Victoria                     42
## 48 Dromana                                0.0111 Victoria                    153
## 49 Kiama Hinterland - Gerringong          0.0110 New South Wales             132
## 50 Hope Island                            0.0110 Queensland                   78
## # ... with 604 more rows

data %>% 
  filter(SA2_NAME16 %in% 
           c("Sydney - Haymarket - The Rocks", 
             "Hobart")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  # geom_point(aes(y = adr_50p_int, size = n_listings_int), 
  #            col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

3.9.2 Means from the individual times

data %>% 
  select(SA2_NAME16, STE_NAME16, date, n_listings_int, diff) %>% 
  arrange(diff) %>% 
  print(n = 50)

## # A tibble: 17,658 x 5
##    SA2_NAME16                   STE_NAME16     date       n_listings_int    diff
##    <chr>                        <chr>          <date>              <int>   <dbl>
##  1 Sydney - Haymarket - The Ro~ New South Wal~ 2019-01-01            209 -0.268 
##  2 Sydney - Haymarket - The Ro~ New South Wal~ 2019-02-01            209 -0.179 
##  3 Flinders                     Victoria       2017-02-01            145 -0.179 
##  4 Sydney - Haymarket - The Ro~ New South Wal~ 2018-12-01            211 -0.178 
##  5 East Melbourne               Victoria       2018-01-01             18 -0.150 
##  6 East Melbourne               Victoria       2017-12-01             18 -0.148 
##  7 West Coast (Tas.)            Tasmania       2018-12-01             69 -0.147 
##  8 East Melbourne               Victoria       2017-11-01             19 -0.144 
##  9 East Melbourne               Victoria       2017-10-01             19 -0.141 
## 10 East Melbourne               Victoria       2017-09-01             18 -0.139 
## 11 Flinders                     Victoria       2017-06-01            193 -0.139 
## 12 West Coast (Tas.)            Tasmania       2018-11-01             69 -0.136 
## 13 East Melbourne               Victoria       2017-08-01             18 -0.136 
## 14 Flinders                     Victoria       2018-07-01            246 -0.135 
## 15 Coolangatta                  Queensland     2017-10-01             28 -0.134 
## 16 Double Bay - Bellevue Hill   New South Wal~ 2017-03-01            117 -0.134 
## 17 West Coast (Tas.)            Tasmania       2019-02-01             69 -0.131 
## 18 Coolangatta                  Queensland     2017-09-01             26 -0.131 
## 19 Coolangatta                  Queensland     2017-02-01             15 -0.130 
## 20 Bathurst Region              New South Wal~ 2017-08-01             29 -0.125 
## 21 Hunters Hill - Woolwich      New South Wal~ 2016-12-01             27 -0.124 
## 22 Coolangatta                  Queensland     2016-12-01             15 -0.123 
## 23 Coolangatta                  Queensland     2017-06-01             25 -0.120 
## 24 Coolangatta                  Queensland     2017-05-01             25 -0.116 
## 25 Jindabyne - Berridale        New South Wal~ 2017-01-01            127 -0.116 
## 26 Sydney - Haymarket - The Ro~ New South Wal~ 2018-11-01            204 -0.115 
## 27 Bathurst Region              New South Wal~ 2018-03-01             34 -0.114 
## 28 Bathurst Region              New South Wal~ 2017-10-01             35 -0.113 
## 29 Jindabyne - Berridale        New South Wal~ 2016-12-01            124 -0.113 
## 30 Broulee - Tomakin            New South Wal~ 2017-01-01             47 -0.112 
## 31 Trigg - North Beach - Water~ Western Austr~ 2017-09-01             32 -0.112 
## 32 Trigg - North Beach - Water~ Western Austr~ 2017-10-01             32 -0.111 
## 33 Bathurst Region              New South Wal~ 2018-02-01             34 -0.110 
## 34 Coolangatta                  Queensland     2017-03-01             20 -0.110 
## 35 Bathurst Region              New South Wal~ 2017-09-01             32 -0.109 
## 36 Broulee - Tomakin            New South Wal~ 2016-12-01             42 -0.108 
## 37 Mosman Park - Peppermint Gr~ Western Austr~ 2018-12-01             34 -0.107 
## 38 Grant                        South Austral~ 2017-05-01             22 -0.106 
## 39 Bathurst Region              New South Wal~ 2018-01-01             35 -0.106 
## 40 Mosman Park - Peppermint Gr~ Western Austr~ 2019-01-01             35 -0.106 
## 41 Queanbeyan Region            New South Wal~ 2017-09-01             18 -0.105 
## 42 City Beach                   Western Austr~ 2017-10-01             17 -0.103 
## 43 Bathurst Region              New South Wal~ 2017-07-01             26 -0.102 
## 44 Queanbeyan Region            New South Wal~ 2017-08-01             18 -0.102 
## 45 Ulladulla Region             New South Wal~ 2017-05-01            154 -0.101 
## 46 Dover Heights                New South Wal~ 2019-02-01             88 -0.101 
## 47 Bathurst Region              New South Wal~ 2018-04-01             36 -0.0990
## 48 Queanbeyan Region            New South Wal~ 2017-07-01             17 -0.0988
## 49 Flinders                     Victoria       2018-08-01            246 -0.0986
## 50 Bathurst Region              New South Wal~ 2017-11-01             35 -0.0985
## # ... with 1.761e+04 more rows

data %>% 
  filter(SA2_NAME16 %in% 
           c("Sydney - Haymarket - The Rocks",
             "East Melbourne")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  geom_point(aes(y = adr_50p_int, size = n_listings_int),
             col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")

data %>% 
  select(SA2_NAME16, STE_NAME16, date, n_listings_int, diff) %>% 
  arrange(desc(diff)) %>% 
  print(n = 30)

## # A tibble: 17,658 x 5
##    SA2_NAME16                     STE_NAME16     date       n_listings_int  diff
##    <chr>                          <chr>          <date>              <int> <dbl>
##  1 Broulee - Tomakin              New South Wal~ 2018-03-01             73 0.159
##  2 Broulee - Tomakin              New South Wal~ 2018-01-01             67 0.141
##  3 Kensington (NSW)               New South Wal~ 2017-12-01             32 0.138
##  4 Jindabyne - Berridale          New South Wal~ 2018-02-01            285 0.124
##  5 Jindabyne - Berridale          New South Wal~ 2018-03-01            300 0.121
##  6 West Coast (Tas.)              Tasmania       2016-12-01             25 0.114
##  7 Broulee - Tomakin              New South Wal~ 2018-02-01             66 0.113
##  8 Sydney - Haymarket - The Rocks New South Wal~ 2017-03-01             37 0.112
##  9 Sydney - Haymarket - The Rocks New South Wal~ 2017-04-01             38 0.112
## 10 Hobart                         Tasmania       2018-04-01            200 0.109
## 11 Broulee - Tomakin              New South Wal~ 2018-04-01             74 0.108
## 12 Rose Bay - Vaucluse - Watsons~ New South Wal~ 2018-02-01            100 0.108
## 13 Sydney - Haymarket - The Rocks New South Wal~ 2017-12-01             56 0.108
## 14 Sydney - Haymarket - The Rocks New South Wal~ 2018-01-01             60 0.107
## 15 Ulladulla Region               New South Wal~ 2018-11-01            290 0.107
## 16 Rose Bay - Vaucluse - Watsons~ New South Wal~ 2018-01-01            102 0.107
## 17 Sydney - Haymarket - The Rocks New South Wal~ 2018-03-01             96 0.107
## 18 Sydney - Haymarket - The Rocks New South Wal~ 2018-04-01            170 0.106
## 19 Bathurst Region                New South Wal~ 2018-08-01             40 0.106
## 20 Sydney - Haymarket - The Rocks New South Wal~ 2018-05-01            173 0.106
## 21 Sydney - Haymarket - The Rocks New South Wal~ 2017-02-01             32 0.106
## 22 Grindelwald - Lanena           Tasmania       2017-11-01             42 0.106
## 23 Jindabyne - Berridale          New South Wal~ 2017-02-01            127 0.105
## 24 Sydney - Haymarket - The Rocks New South Wal~ 2018-07-01            187 0.105
## 25 Sydney - Haymarket - The Rocks New South Wal~ 2018-08-01            189 0.105
## 26 Coolangatta                    Queensland     2019-01-01             53 0.105
## 27 Sydney - Haymarket - The Rocks New South Wal~ 2018-09-01            199 0.104
## 28 Ulladulla Region               New South Wal~ 2018-10-01            275 0.104
## 29 West Coast (Tas.)              Tasmania       2017-01-01             27 0.103
## 30 Galston - Laughtondale         New South Wal~ 2017-10-01             19 0.103
## # ... with 1.763e+04 more rows

data %>% 
  filter(SA2_NAME16 %in% 
           c("Broulee - Tomakin", 
             "Rose Bay - Vaucluse - Watsons Bay")) %>% 
  ggplot(aes(x = date)) +
  geom_line(aes(y = m5_fit, group = SA2_NAME16), 
            col = "darkorchid4") +
  geom_point(aes(y = rent_50p_int, size = record_count_int), 
             col = "#00AFBB", alpha = 0.3) +
  geom_point(aes(y = adr_50p_int, size = n_listings_int),
             col = "#E7B800", alpha = 0.33) +
  theme_minimal() + 
  xlab("") + ylab("Rent / fitted") + 
  facet_wrap(~SA2_NAME16, scales = "free_y")